Taking PayPal private — what price the Institute believes the board requires, and why the money never reached it

The Institute believes the board will not sell below $70, and that the answer is $70 or higher. The money that was actually assembled reached $64.77. That is the whole case.

Everything else in a take-private analysis — the leverage, the tax shield, the cost program, the exit multiple — is academic until a buyer and a seller agree on a price. A price being stated is not enough: Advent International and Stripe stated $60.50 on July 15, 2026 and the sellers declined it. So this case starts where the negotiation actually starts. Four methods put the price the Institute believes the board requires at roughly $70 a share — and two cautions travel with that, stated here rather than buried. The discounted cash flow’s terminal value implies an exit multiple of 9.33 times, which is essentially the 9.47 times peer median, so methods three and four are one method twice. And the buyback average is produced by a fixed annual dollar budget rather than by directors choosing a price. The conclusion survives both; the claim of four independent confirmations does not. Advent International and Stripe indicated $60.50 on July 15, 2026, were told the price was inadequate, and withdrew on August 28, 2026. They were about $5.23 a share short — roughly $4.6 billion of equity.

A reservation price is an inference, not a disclosure — no company files one and PayPal has not stated one. What can be observed is the price the board itself paid with shareholder money, the price the public market paid before any bid existed, what the peer group is worth on the same trailing statistics, and what the cash flows support on their own. The buyback average is the fiscal 2025 figure, the 52-week high was struck October 28, 2025 — nine months before the approach became public, so no takeover speculation is inside it — and the discounted cash flow is derived in full below.

Against that, the funding that actually existed. Roughly $50 billion of committed debt from J.P. Morgan and Morgan Stanley plus $17.0 billion of equity from Stripe, Advent and Block clears total uses at $64.77 a share. That is the ceiling the assembled package reached, and it is real: two of the largest lenders in the world quoted the debt against these cash flows, and the equity was subscribed in weeks. Here is where the number comes from.

Three prices, three different things. $60.50 is what Advent and Stripe offered and the sellers declined. $64.77 is the ceiling of the money they had actually assembled — what they could have stretched to had they chosen. $70 is what the Institute believes the seller requires. The bid was $3.73 below the buyer’s own ceiling, which says the sponsors were not bidding to the limit of their funding; the ceiling was still $5.23 short of the price the Institute believes gets a deal done.

The gap is not a valuation disagreement. It is an equity-assembly problem. Fund the Institute’s $70 with more debt and total borrowings reach $54.6 billion — 7.91 times trailing EBITDA, with 49.4% of the interest bill pushed past the Section 163(j) line into deferral. (At the banks’ committed $50 billion alone the figures are 7.24 times and 44.7%; the higher pair is what funding the whole gap with debt requires.) Fund it with more equity and the check becomes $21.6 billion — 1.31 times what was actually raised, and 62% of the roughly $36 billion that closed the Electronic Arts take-private three weeks earlier, of which about 93 percent came from a single sovereign fund. Both jaws close at the same time.

What follows works that through in seven steps: the Institute’s estimate, the discounted cash flow behind it, the debt that has to be refinanced, the sources and uses at three prices, the equity check measured against checks anyone has actually written, what borrowing more and cutting more can and cannot do to the gap, and the one structure that clears both sides.

Disclosure. The author holds no position in PayPal Holdings, Inc., no position in any other security named in this case, and no economic interest in any of the transactions discussed. This case is educational and reflects the author’s research framework. It is not investment advice and is not a recommendation to buy, sell, or hold any security. The author is a CPA and MBA publishing under the Baratelli Institute and is not a registered investment adviser.

A board that spent $20.2 billion buying its own stock left a record — but it is a record of a budget, not of a price judgment

This is the method a director raises first, and it is the one this section will end up trusting least. Every other method asks what the company is worth to somebody. This one asks what the company itself paid, in cash: between fiscal 2023 and the second quarter of 2026 PayPal retired 319 million shares for $20,179 million.

Computed from the consolidated statements of stockholders’ equity in the fiscal 2025 Form 10-K and the Form 10-Q for the quarter ended June 30, 2026. Average price is dollars divided by shares within each period as labeled; the whole-period average is the aggregate, not an average of averages.

The 2026 rows have to be handled honestly rather than dropped. PayPal repurchased at an average of $44.50 in the first quarter and $45.88 in the second, which is well below $70 and looks at first like the board marking its own view down. It is not. The 52-week low of $38.46 was struck on February 12, 2026, inside that first quarter. A buyer executing a standing authorization at $44.50 while the stock prints $38.46 is buying a dislocation, not marking a view down — but neither is it marking one up. The board set the budget; the tape set both prices. The fiscal 2025 average of $70.38 — struck across a full year rather than one drawdown — is the cleaner read, and it lands within forty cents of $70.

What this method cannot claim. The average price in that table is not the board’s number — it is the market’s. A board does not set a repurchase price. It approves an authorization: a dollar amount and a window. The shares then get bought at whatever price other market participants happen to be transacting at while that money is spent, so the change in the average is driven by market participants, not by any change in the directors’ view of value. Look down the dollar column and the mechanic is visible. Fiscal 2024 and fiscal 2025 are $6,053 million each — the same number twice. The first two quarters of 2026 are $1,513 million and $1,514 million. Management’s own 2026 guidance is approximately $6,000 million. That is not a board opportunistically pricing its stock; it is a fixed annual dollar budget spent down on a schedule, and under a fixed-dollar program the average price paid is not chosen by anybody — it is mechanically the average price the stock traded at over the spending window. So $70.38 is closer to a description of where PayPal traded in fiscal 2025 than to a statement of what its directors think the company is worth. Two things survive that, and only two. The board kept funding the program, and guided to funding it again — directors who thought their stock expensive would cut the authorization instead. And the fixed-dollar mechanic actually strengthens the 2026 rebuttal: if price is not the decision variable, the $45.18 paid during the February collapse is not a markdown of the board’s view either. This is the weakest of the four methods, and it is listed first only because it is the one a director will raise first.

The second read: the highest price the public market ever paid, struck before any bid existed

PayPal traded at $79.22 on October 28, 2025. That matters because of when it happened: nine months before the Advent and Stripe approach became public on July 15, 2026. There is no takeover speculation inside it. It is what the marginal public buyer paid for the business on its own merits, and a board negotiating a sale in 2026 knows its holders saw that print within the year. Its weakness is that it is a single tick rather than a distribution — which is why it is one of four methods and not the method.

The third read: what the peer group is worth on the same trailing statistics

Prices and trailing statistics from S&P Global Market Intelligence, retrieved August 30, 2026. The median of the three non-PayPal peers is 9.47 times. PayPal trades at 7.57 times on the same basis, a discount to a group it is larger than by revenue.

One caution belongs on the median itself. Fiserv at 7.20 times is not a neutral yardstick here. It trades at 7.20 times because it has a drawdown of its own, which is to say the peer read is anchored on a second depressed payments multiple rather than a healthy one. That cuts both ways, and the Institute prints both. The version that strengthens the argument is that a sector-wide de-rating makes it more likely a whole category of cash generators is mispriced rather than that PayPal alone has a broken business — which is the case for taking one of them private. The version that weakens it is the mirror image: a median built out of dislocated marks reproduces the dislocation, so the $68.04 that method returns is the softest of the four reads, and it is the lowest of the four for exactly that reason. Read alongside the Institute’s Fiserv case, which works the same compression from the inside of that company.

On the median non-PayPal peer multiple of 9.47 times trailing EBITDA, PayPal is worth $68.04 a share after the net-debt bridge. That figure is struck on the vendor’s trailing EBITDA of $6,460 million, the basis every multiple in the table above uses; it differs from this case’s $6,909 million because of period cut-off and add-back conventions, and on the filings’ basis the same multiple implies $73.01. The lower of the two is carried here. Peer multiples on August 28, 2026 closes, trailing statistics on the same basis PayPal is measured. The board does not need to believe PayPal deserves a premium to the group to believe it deserves the group’s own multiple.

The spread is $68 to $79, and the convergence inside it is real but narrower than four rows suggest. The buyback average ($70.38) and the peer read ($68.04) land $2.34 apart and are genuinely independent of each other — but the first is a budget artifact, as the subsection above shows, and the second is the same multiple the discounted cash flow assumes on its way out. What is left standing is one method with real force, one corroboration of it, and a 52-week high ($79.22) that is a negotiating anchor rather than a valuation. Roughly $70 is the conservative reading of that evidence, not the aggressive one — it sits below the midpoint of the four, and it is the number this case uses everywhere after.

PayPal is a cash gusher, and the cash flows on their own say $76.03

The three reads above are all relative — to what the board paid, to what the market paid, to what peers trade at. A discounted cash flow is the one method that does not reference a price at all, which is exactly why it belongs here. Two conventions govern it and both are stated before any number appears, because both cut against the answer.

Stock compensation is not added back. PayPal recorded $1,001 million of stock-based compensation over the trailing twelve months to June 30, 2026. It is already inside GAAP operating income and this case leaves it there. Adding it back would raise the valuation by treating a real transfer to employees as free cash. Working capital is held flat. Funds receivable of $39,743 million and funds payable of $41,743 million are customer float; they move together, and treating their swing as an owner’s cash flow would be an error in either direction.

Computed from the fiscal 2025 Form 10-K and the Form 10-Q for the quarter ended June 30, 2026. The 20 percent rate allows for state tax against a reported non-GAAP effective rate of 17.1 percent in the June quarter. This unlevered figure sits below the $6,586 million of levered free cash flow the company generated over the same period and below management’s “$6 billion plus” adjusted free cash flow guide, because it is taxed at the full corporate rate with no interest shield and no stock-compensation add-back.

The discount rate is built, not asserted — and it runs on its own clock

Three clocks run in this table and none is blended: the risk-free rate and the beta at September 1, 2026, the equity weight at the August 28, 2026 close, and the debt weight from the June 30, 2026 balance sheet. The equity risk premium and the cost of debt are Institute assumptions rather than quoted terms and are the two inputs a reader is most entitled to disagree with.

The result, and the cross-check that keeps it honest

No cost savings are in this number. PayPal has announced at least $1.5 billion of run-rate cost reductions and had delivered roughly $400 million of it on a July 2026 clock. None of it is credited here. The unlevered free cash flow below is built from operating income the company has already reported, so every dollar the program eventually delivers is upside to this discounted cash flow rather than an input to it. Section 6 works the cost program separately, where it belongs, because savings that have not happened yet are a financing argument rather than a valuation input.

Growing unlevered free cash flow at 3% for five years and 2% thereafter, discounted at 9.64%, produces an enterprise value of $67,270 million. Bridging to equity at the $2,228 million of net debt on the corporate cash actually available gives $65,042 million, or $76.03 a share. On the alternative bridge that counts every dollar of cash and investments on the balance sheet, net debt is negative and the value is higher; the conservative basis is used.

The terminal value is 70% of the total, which is normal for a five-year explicit period and is also the reason a cross-check is required. A single-stage Gordon growth model on the same cash flow and the same rate lands at $72.76 — below the two-stage result, which is what should happen when the two-stage model is given a higher first-stage growth rate. If the cross-check landed above, the model would be broken.

The disclosure most models leave out: what that terminal value is actually assuming

When 70 percent of enterprise value sits in the terminal, the Gordon formula is not a neutral continuation assumption. It is a disguised exit multiple, and it should be stated as one. Terminal-year EBITDA is $8,009 million; the $74,739 million terminal value is therefore 9.33 times that figure, or 10.82 times trailing EBITDA. Compare that to the 9.47 times median of the peer group used in the section above and the two are essentially the same number — which means the discounted cash flow and the comparables are not two independent reads. They are one assumption arriving twice by different routes.

Set against where PayPal actually trades, the same discipline cuts the other way. On this case’s net-debt basis of $2,228 million, PayPal changes hands at 6.97 times trailing EBITDA; on the alternative basis that credits every investment on the balance sheet, 6.39 times. Neither is wrong; quoting one against a multiple computed on the other is. Substitute the traded multiple for the Gordon terminal and the same cash flows are worth $62.08 a share rather than $76.03. Solve the model backwards and $70 requires an exit multiple of 8.31 times — above where PayPal trades today, below the peer median, and squarely inside the range a board can defend.

So the honest statement of what this method produces is conditional. $76.03 is the value if PayPal re-rates toward its sector. $62.08 is the value if it never does. That is why the Institute reads $70 as a floor the board can defend rather than as a valuation the market currently supports.

Each cell is the full two-stage model rerun at that rate and that terminal growth, on the same unlevered cash flow, the same five-year explicit period and the same net-debt bridge. Nothing else is changed.

21 of the 30 cells exceed $70 a share. Every one of the nine that does not requires a discount rate of 10.0 percent or higher — above PayPal’s own computed cost of capital of 9.64 percent — paired with terminal growth of 2.5 percent or less. Neither assumption is absurd; both are pessimistic relative to what the inputs support. At 2.0 percent terminal growth the two rows that bracket the computed cost of capital return $77.54 at 9.5 percent and $72.50 at 10.0 percent — both above the Institute’s estimate. A board looking at this grid does not conclude that $60.50 is a fair price. It concludes that $60.50 is a discount to almost every reasonable parameterization of its own cash flows, which is the finding, and it is consistent with the three relative methods in the section above.

The change-of-control put is a funded refinancing, not a wall — and it costs about $460 million a year

PayPal carried $12,728 million of term notes at June 30, 2026 across seventeen series at a principal-weighted effective rate of 4.15 percent. The near-term ladder is comfortable: $3,528 million falls due over the thirty months from the June 30, 2026 balance sheet date to the end of 2028 against trailing free cash flow of $6,586 million a year. Nothing forces a transaction, and a board under no refinancing pressure is a board that can hold out on price.

Note 12 provides that upon both a change of control and a downgrade below investment grade, PayPal must offer to repurchase every series at 101 percent of principal plus accrued interest. A take-private satisfies the first by definition and one financed at buyout leverage satisfies the second with certainty, so all $12,728 million becomes putable at $12,855 million. A holder of 2.300 percent notes due 2030 tenders instantly, because 101 is far above where that paper trades.

That is a cost line rather than an obstacle. The roughly $50 billion of debt reportedly committed by J.P. Morgan and Morgan Stanley in July 2026 is roughly four times the size of the put — the suitors who actually ran at this company had already funded it. What survives is the spread: replacing 4.15 percent paper with market-rate paper costs on the order of $460 million a year of incremental interest at a 7.0 to 8.5 percent replacement rate — $426 million at the 7.5 percent blended coupon this case assumes throughout. Real money, and small against a six-dollar-a-share price gap.

Sources and uses at three prices

Share count is the diluted weighted average for the three months ended June 30, 2026. Fees are an Institute assumption built by component rather than as one rate on the whole deal: financing fees of 1.75 percent on the debt actually raised, advisory of 0.40 percent on the purchase price, and $160 million for legal, accounting, regulatory filings and D&O run-off. An earlier version charged a single 2.5 percent rate and produced $1,865 million at $70 — roughly forty-five percent high, because it charged a financing rate on equity that raises no financing. The $47.27 unaffected price implied by the reported 28 percent premium is deliberately not run as a column: nobody buys this company at the level it traded before a bid existed, so a column of arithmetic there would settle nothing. It is used once, in the premium calculation below, which is the one job it can do. The $75.00 column is here because $70 is an estimate of a seller’s floor, not a ceiling — a board declining $70 may be holding for more.

One number belongs next to the Institute’s estimate and is easy to skip past. At $70.00 against the $47.27 unaffected price the premium is 48.1 percent, and at $75.00 it is 58.7 percent — against the 28 percent actually indicated, and well above the premium in the Electronic Arts transaction. That does not make $70 unreasonable, but it is the number a buyer’s investment committee sees first, and it is part of why the gap was not simply split.

The ceiling also moves with the debt, and it is worth seeing how far. The $64.77 above is struck on the banks’ committed $50 billion. At $45 billion the funding reaches $59.22; at $55 billion it reaches $70.32 — roughly $1.11 a share for every additional billion, and at $55 billion the gap to the Institute’s estimate essentially disappears. That $55 billion was not on the table because 7.91 times leverage is not what these lenders were quoting against declining trailing operating income; the sensitivity shows the constraint is the size of the debt package, not the arithmetic of the sources and uses.

On the sources side, PayPal reports $15,265 million of cash and investments, but only about $5,000 million of it is spendable by a buyer at closing. The rest is customer money: amounts due to customers of $41,743 million exceed segregated funds receivable, and regulatory capital held against those balances is not available to finance an acquisition. Treating the headline cash number as deal funding is the single most common error in a payments buyout, and against the $15,265 million balance-sheet figure it is worth about $10 billion — nearly $17 billion if the reader reaches instead for the $21,973 million of cash, cash equivalents and restricted cash the cash-flow statement ends the period on.

Equity is the residual at every debt level: total uses less new debt less the $5,000 million of corporate cash. Trailing EBITDA of $6,909 million for the twelve months ended June 30, 2026.

Read across that bottom set of rows and the structure of the problem is visible. Every extra dollar of price is a dollar of equity, one for one, because the debt is already at its committed maximum. Moving from $60.50 to $70.00 adds $8.6 billion of uses and every cent of it lands on the equity check. Price risk in this transaction is equity risk.

$21.6 billion is not unraisable. It is unassembled — and that is a different problem.

The easy answer to “can anyone write $21.6 billion of equity?” is no, and the easy answer is wrong. Three weeks before Advent and Stripe walked away, roughly $36 billion of equity closed into a single take-private: Electronic Arts, acquired for $55 billion by a group led by Saudi Arabia’s Public Investment Fund, announced September 29, 2025 and completed August 4, 2026 — the largest all-cash sponsor take-private on record, and 65 percent equity. The money exists, at this scale, on this clock.

The structure of that transaction is the instructive part. Of the roughly $36 billion, the Public Investment Fund took the balance of roughly 93 percent, the private equity sponsor 5.5 percent — about $2 billion — and the third partner 1.1 percent. The most active large-cap technology take-private sponsor in the world wrote a two-billion-dollar check into the largest buyout ever done, and a sovereign wealth fund wrote the rest. A large-cap fund runs $15 to $25 billion of commitments and, under concentration policies most funds cap between 15 and 25 percent, a single-name position of $3 to $6 billion is the practical ceiling. Equity at this size is assembled, not allocated, and the anchor is sovereign or strategic.

Who could write it — demonstrated capacity, not expressed interest

None of the parties below has said anything at all about PayPal. They appear because equity of this size is raised through a conversation with somebody who has already signed a check that large, and PayPal is unusually well connected to that list — its founding generation built several of these institutions.

Capacity figures only, carrying the dates shown, and third-party reporting rather than filed disclosure except the Electronic Arts structure, which is a completed transaction. The Advent figure is the firm’s own published total at March 31, 2026; some January 2026 announcements cite a figure above $100 billion and the smaller, later measurement is used. The Stripe valuation is the price struck in its own February 24, 2026 employee tender — a real transaction in the shares, but a private mark rather than a quoted price. A net worth is a mark on concentrated listed stock rather than a cash balance. The single-fund ceiling is an Institute estimate of standard concentration practice, not a disclosed limit at any firm.

The combination the Institute would put on the list, and nobody has

Michael Dell’s family office, DFO Management, alongside Silver Lake. This is not a rumor and no approach of any kind has been reported; it is a capacity and experience argument, and it is the one pairing on this page with a completed precedent at the exact structure PayPal requires. The 2013 Dell take-private was not a sponsor buying a company. It was a founder rolling his own shares as the anchor equity while a single sponsor wrote alongside him, and it is the only benchmark in the table below where the anchor check came from one person rather than a fund. Silver Lake was the sponsor in 2013, has been beside Dell ever since, and took 5.5 percent of the Electronic Arts equity in August 2026 — so the pairing is not historical, it is current. What followed the 2013 deal is the part that matters more than the deal itself: EMC in 2016 at roughly $67 billion, the largest technology acquisition ever attempted at that point, financed and then deleveraged, with the company returned to the public market in 2018. That is a demonstrated record of buying a large, unloved technology business, carrying an uncomfortable debt load through it, and taking cost out without destroying the asset. It is precisely the skill this transaction needs, and precisely the skill a generalist buyout fund does not have. Put plainly: Michael Dell has already run this transaction once, at larger scale and under worse conditions. He would be doing it again, not doing it for the first time.

Where the money would come from, and where it would not. This is not Dell Technologies buying PayPal. The public company would not be a party, its balance sheet would not be used, and its shareholders would not vote. It is a leveraged buyout run off the private side: DFO Management, the family office that manages Michael Dell’s own capital, writes the anchor equity; Silver Lake takes the sponsor seat it has held beside Dell since 2013; and equity beyond what those two want to hold is syndicated through BDT & MSD Partners, the merchant bank Byron Trott’s BDT & Company formed with MSD Partners in January 2023, whose whole business is raising capital from family- and founder-led investors. That firm is a separate house from DFO rather than a captive checkbook, but it points at exactly the pool of money this deal needs — which is why a family office can credibly reach for a $21.6 billion check without writing all of it. What that structure cannot do is the thing the Musk companies do. Cooperation between a Dell-anchored PayPal and Dell Technologies is attractive on paper — shared engineering, one commercial direction — but Dell Technologies is public, with independent directors and duties running to minority holders, and it cannot be pointed at a related private transaction the way a wholly owned entity can. The cooperation thesis is real at the family-office level and constrained at the public-company level.

The question underneath the Dell thesis is the right one to ask: is PayPal essentially a technology company? Substantially, yes — it is a software and platform business with 2.6 percent capital intensity, and Section 2 shows a cost structure that behaves like one. But it is also a licensed money transmitter in every state and a lender through its credit book, and that regulatory overlay is the honest difference from Dell in 2013. A hardware company changes hands with an antitrust review. A money transmitter changes hands one state approval at a time, and Section 7 puts that timeline at twelve to eighteen months. The technology thesis is right about what the asset is; it understates what it takes to close on it.

Now the honest caveat: capacity is not willingness, and a mandate is a tighter constraint than a balance sheet. Founders Fund’s limited partners subscribed for venture risk, not for a levered payments buyout, and the same objection applies to Andreessen Horowitz and to Thrive Capital. A sovereign fund that has just deployed thirty-four billion dollars into one asset is not obviously reaching for another. And a strategic paying in its own stock is spending an asset its own holders price, which is a different negotiation entirely.

Where PayPal is structurally worse off than the precedents

Enterprise values as reported at announcement. The Dell 2013 row is the completed transaction at $13.75 per share plus a $0.13 special dividend; its funding was $3.7 billion of rolled founder shares, $0.5 billion of family-office cash and $1.4 billion of sponsor equity against $13.8 billion of new term loans and notes and Microsoft’s $2.0 billion subordinated loan, with a further $3.5 billion from Dell’s own cash and receivables financing — which is why equity and debt do not sum to the transaction value. Those figures are reconstructed from filings and contemporaneous press rather than a disclosed sources-and-uses table. TXU is the cautionary case: an 18 percent equity contribution against cash flows that moved against the sponsors, Chapter 11 in 2014, equity wiped out.

The Dell row carries the one structural advantage PayPal does not have. Of that $5.6 billion of equity, $3.7 billion was Michael Dell’s own stock, rolled rather than bought — the genuinely new cash was about $1.9 billion against a $24.9 billion transaction, under eight percent. A founder with a large rollable stake is the cheapest equity in a buyout, because his shares reduce what has to be purchased instead of funding the purchase. PayPal has no such holder: it was spun out of eBay in July 2015 with an institutional register and no founder position, so every dollar of the check above has to arrive as new money on closing day.

State the finding at the right width. The equity is not unraisable — it is unassembled. The July 2026 group reached $17.0 billion of the $21.6 billion the Institute’s estimate requires, with three participants and no anchor. The remaining $4.6 billion is one more seat at a table that has already been built once, and a sovereign, a large strategic or a rollover from existing holders could each supply it alone. What the arithmetic rules out is a sponsor-only club clearing $70 a share. It does not rule out the transaction.

Two objections arrive here. Each moves a genuine constraint. Neither moves the one that decided this.

Borrow more: what the tax code defers rather than denies

Section 163(j) limits the deduction for business interest to 30 percent of adjusted taxable income. The One Big Beautiful Bill Act, signed July 4, 2025, permanently restored the EBITDA-style measure for tax years beginning after December 31, 2024 — depreciation and amortization are added back again, which made large buyouts materially more financeable. Any analysis still running the EBIT basis is on a stale clock.

On the current basis, trailing tax EBITDA of $6,909 million supports $2,073 million of currently deductible interest, which at a 7.5 percent blended coupon shelters about $27,636 million of debt — four times EBITDA. Above that line the interest is not permanently disallowed and this is not a borrowing limit: under §163(j)(2) the excess carries forward indefinitely for a C corporation. It is a timing cost, not a forfeiture.

Cash tax applies a 20 percent blended rate to EBITDA less depreciation and amortization less deductible interest, against a reported non-GAAP effective rate of 17.1 percent in the quarter ended June 30, 2026. Capital expenditure is held at the trailing $889 million against company guidance of approximately $1.0 billion, so the last column is if anything favorable. No growth, no cost program and no working capital release is assumed anywhere in this table. Adjusted taxable income is approximated by EBITDA, which is the correct basis after December 31, 2024; the approximation is generous in one further respect, since for tax years beginning after December 31, 2025 the Act also excludes certain foreign inclusions from ATI, and PayPal earns a substantial share of its income outside the United States.

The committed row answers the first objection. At $50 billion the structure is already 7.2 times EBITDA and 44.7% of the interest bill is deferred in the year it is paid. The business throws off $1,495 million after capital expenditure, interest and tax — three percent of the debt stack. Borrowing the extra $4.6 billion to reach $70 takes leverage to 7.91 times and the deferred share to 49.4 percent, on a business whose trailing operating income is declining. The lenders are not the constraint. The point at which the tax code and the cash flow both stop cooperating is.

Cut more: the one lever with real force, and its limit

A cost program does double duty, because 30 cents of every permanent dollar removed becomes deductible interest capacity — four dollars of sheltered debt at a 7.5 percent coupon. Four to one. Nothing else inside this structure has that leverage, which is why a sponsor pursues cost well past the point public management will commit to publicly. PayPal employed approximately 23,800 people at December 31, 2025, and slide 7 of its second-quarter 2026 presentation sets out at least $1.5 billion of gross run-rate savings phased through 2029. Those carry a July 2026 clock and the headcount a December 31, 2025 one; neither is an achieved result.

Sheltered debt is 30 percent of the case EBITDA divided by a 7.5 percent coupon. The addressable non-transaction pool — customer support, sales and marketing, technology, and general and administrative — annualizes to $9,216 million from the six months ended June 30, 2026. The sponsor case is an underwriting assumption made to size the requirement, not a forecast and not a prediction of headcount.

But the shelter is not the check. Removing a quarter of the addressable cost pool, eliminating stock compensation entirely and stripping out public-company costs takes sheltered debt from $27,636 million to $41,056 million. That relieves the tax constraint, improves the returns and lifts the credit. It hands the buyer no additional dollar to spend, which is why the tax channel never closes the gap on its own. The channel that does is next.

The channel that actually moves the price — how lenders size the program

The tax code is the smaller of the two channels. The larger one is the lenders, who size a commitment off a leverage multiple rather than off a tax section — and credit agreements routinely let a borrower run that multiple against run-rate EBITDA including announced synergies rather than trailing EBITDA as reported.

The multiple is not assumed. It is $50,000 million of committed debt divided by the $6,909 million of trailing EBITDA the banks were underwriting against — 7.237 times — held constant across both columns so the only thing changing is the EBITDA denominator. The equity column is unchanged at $17,000 million throughout; this table adds no equity. The run-rate column adds the company’s own $1,500 million program and the $50 million of public-company cost a private owner stops paying — $1,550 million in total, 22.4 percent of trailing EBITDA, which sits at the upper end of, but inside, the synergy add-back caps typically negotiated into credit agreements.

That is the honest second branch of this case. On trailing EBITDA the assembled funding stops at $64.77 and $70 is out of reach. On run-rate EBITDA at the same multiple the banks already accepted, the commitment goes to $61.2 billion, the affordable price to $77.21, and the equity check at $70.00 falls to $10.6 billion — inside the $17.0 billion these buyers had already raised, with room. $75.00 clears as well.

So the question is a credit question rather than a valuation one: will a bank size $61.2 billion against a program that has delivered roughly $400 million of a $1,500 million target on a July 2026 clock? That is a materially softer underwrite than trailing EBITDA — execution risk on cuts at a company whose branded checkout is losing share — and a bank that says no is not being unreasonable. The July 2026 group stopped at $50 billion and $60.50, which is the behavior of buyers not given the run-rate basis: evidence about how this financing was sized, not proof of what any financing must look like. Everything else on this page runs on the trailing basis, which is the conservative one.

Three constraints move and the fourth cannot. The debt §163(j) fully shelters goes from $27.6 billion to $41.1 billion; the returns improve; the credit lifts. The equity actually assembled stays at $17.0 billion, because a cost program does not enlarge anyone’s balance sheet.

Does PayPal bring a loss carryforward to the table?

It is the obvious question at this point in a leveraged structure, and the answer is no. A buyer acquiring a company with a large net operating loss inherits a shield that costs nothing to create. PayPal does not have one. Note 16 to the fiscal 2025 Form 10-K, measured at December 31, 2025, discloses no federal net operating loss carryforward at all. What it does disclose is smaller and mostly unusable in this transaction.

FY2025 Form 10-K, Note 16 — Income Taxes, measured at December 31, 2025. The company took a $496 million valuation allowance increase in 2025, which is management’s own statement that it does not expect to use a portion of these attributes. Amounts are as filed and would be subject to whatever limitation the transaction itself imposes.

So the interest deduction worked in this section is the only tax lever the transaction has. There is nothing inherited to fall back on, and the $299 million capital loss shelters gains rather than operating income, so it does nothing for an operating company’s tax bill. Two related items belong beside it. PayPal holds a Singapore tax incentive running through 2030 that saved $96 million in 2025 against $473 million in 2024 — a benefit already declining, and one a change of control could put in play. And gross unrecognized tax benefits stood at $2,545 million with a further $637 million of accrued interest and penalties, with examinations open in the United States, California, India, Singapore and Israel back to 2013. In a purchase agreement that is a diligence and indemnity problem, not a valuation adjustment, but it is not nothing.

Every constraint in this case is a constraint on leverage. A buyer paying in stock meets none of them.

The Institute believes the board will not sell below $70, and that the answer is $70 or higher — a reading of the evidence, not a report of anyone’s intention. The funding actually assembled reached $64.77. Everything between those two facts has been worked: the asset underwrites, the debt amortizes, the change-of-control put is funded rather than fatal, the tax code prices leverage rather than forbidding it, and a real cost program moves three of the four constraints. The transaction died on the fourth — the size of the cash equity check — and no structuring inside a leveraged buyout makes that number smaller.

But that is a constraint on financial buyers, and the reason is mechanical: a financial buyer writes a check, a large public company issues shares. What follows is worked as an illustration only. No such transaction has been proposed by anyone, and the Institute expresses no view on whether one will be. Take Space Exploration Technologies, which listed on Nasdaq on June 12, 2026 under SPCX at $135, closed its first session at $160.95 and about $2.1 trillion of capitalization, and closed at $140.95 on August 28, 2026 — the same date this case prices PayPal.

The financial column is the sources and uses above at $70.00. The strategic column assumes the acquirer’s ratings support leaving PayPal’s existing notes in place, which is what defeats the second trigger in the Note 12 repurchase provision; if a rating agency downgraded the notes anyway, the put fires and the strategic column carries the same $12,855 million. That is a ratings judgment, not an arithmetic one. The acquirer share count is now filed: 7,607 million Class A plus 5,569 million Class B, 13,176 million in total at June 30, 2026, priced at the $140.95 close of August 28, 2026. That supersedes the derived count of approximately 13.0 billion used in the first edition of this case — the $2.1 trillion first-day capitalization divided by the $160.95 first-day close — which is retained here and labeled rather than silently swapped.

Read the two bold rows together. The obstacle this case has spent six sections documenting — a $21.6 billion equity check the sponsor market cannot assemble without a sovereign anchor — becomes 3.2% of an acquirer’s share count. And fourteen billion dollars of uses disappears, not negotiated away but never incurred, because the covenant that creates it is conditional on a downgrade a strategic buyer does not trigger. With no acquisition debt there is no interest, no deferred deduction and no deleveraging schedule.

The last question is what PayPal does to the acquirer’s own earnings, because a 3.2 percent dilution is only interesting against what is bought with it.

SpaceX reported adjusted EBITDA of $1,127 million in the first quarter of 2026 and $3,538 million in the second, $4,665 million for the six months ended June 30, 2026. Both annualizations are Institute constructions rather than filed figures and neither is a forecast: the sequential jump from $1,127 million to $3,538 million is large enough that the two bases bracket rather than agree, and the spread between them is itself the finding. Capital expenditure of $28,476 million in the same six months against $4,665 million of EBITDA is the reason unlevered cash flow is worth more to this acquirer than to a sponsor. PayPal EBITDA is the trailing twelve months to June 30, 2026 with no cost program assumed.

That is the trade in one line. Roughly three percent of the acquirer’s equity buys somewhere between a half and three quarters more EBITDA than it currently generates, and it arrives as unlevered cash at a company spending six times its EBITDA on capital equipment. Whether that is a good trade depends on the industrial case, which is where it gets thin.

Two ways to pay, and they are not equally easy. A stock deal need not mean handing PayPal holders SpaceX shares. SpaceX could sell stock into the market and pay cash instead, and the dilution is identical either way — 3.2% of the share count. The execution is not. A share exchange needs no buyer: the shares go to the sellers and the market never has to clear them. Selling for cash asks the market to absorb $61.7 billion of new stock in one marketed transaction, from a company listed only since June 12, 2026, when the largest single-company equity offerings on record sit near twenty billion dollars. Same ownership either way; only one route has to be underwritten.

The industrial case, and where it is thin

What PayPal owns and a new entrant cannot buy in a hurry is the regulatory perimeter: money-transmitter licenses in essentially every American state, a Luxembourg banking license that passports across the European Union, regulated entities in dozens of jurisdictions, a merchant-acquiring stack in Braintree and a consumer network in Venmo. Buying PayPal is buying twenty-five years of licensing and compliance infrastructure at 8.7 times EBITDA at the Institute’s estimate, 6.4 times where the stock actually trades; building it takes a decade and the outcome is not certain. The founder history is a good story and worth nothing analytically — boards do not approve sixty billion dollars of stock issuance on sentiment.

Three objections are serious and none is answered here. Governance: an all-stock deal involving parties with overlapping histories invites entire-fairness review in Delaware and a special committee with independent advisers. Regulatory: a payments combination of this size draws antitrust and prudential scrutiny in every jurisdiction where PayPal holds a license, and license transfers are not automatic. The currency: a buyer paying in its own stock is spending an asset priced by its own holders, and if that price is wrong the arbitrage runs backwards.

The finding. PayPal at the $53.66 close of August 28, 2026 is cheap against every one of the four reads assembled here — four methods, of which two are one method twice, all landing between $68 and $79. The gap between $64.77 and $70 is not a valuation dispute — it is $4.6 billion of unassembled equity, and the constraint is on who is shopping rather than on what the asset is. A withdrawal over price is the most reversible kind there is: no agreement was signed, no standstill has been reported, and the financing was committed against cash flows that have not changed.

PayPal’s own 10-K cover page says the company is worth $71.3 billion. It traded at $45.9 billion.

Figures are dated to when they were measured, which is not always when this case was written. The $71.3 billion aggregate non-affiliate equity value is correctly filed and correctly cited — and it was measured on June 30, 2025, fourteen months earlier, at a level 55 percent above where PayPal actually traded on August 28, 2026.

Balance-sheet items are measured at June 30, 2026 and the trailing figures over the twelve months to that date. The cost program is a forward commitment rather than an achieved result. Market-data inputs are dated September 1, 2026; market prices are August 28, 2026 closes. The transaction rows are not about PayPal; they are analogies.

Material limitations

The offer price, the premium, the financing terms and the identity of the parties in the July 2026 approach are press reporting rather than filed disclosure, because no merger agreement was signed and no proxy was filed; they are labeled as reported wherever they appear. The 7.5 percent blended coupon, the 20 percent cash tax rate, the 5.00% equity risk premium and the component fee stack are Institute assumptions, not quoted terms. The reservation price of roughly $70 is an inference from four methods, not a disclosure — no company files its reservation price and PayPal has not stated one. Cost reduction is modeled as a ladder of scenarios rather than a forecast, and the sponsor case in particular is an underwriting assumption made to size the requirement. No revenue synergies, dividend recapitalizations or asset sales are modeled, and a real buyer would model all three — a sale of the credit book or Braintree would change the uses meaningfully and is not attempted here. Regulatory capital held against customer funds is treated as unavailable, which is the conservative treatment. Reasonable analysts will disagree with several of these assumptions.